However, while the XML Tools pretty-print does work with individual selections, I had forgotten you said there may be thousands, which means it won’t be very efficient. I think the three-step search+replace/xml tools/search+replace is probably the best in-Notepad++ option.

However, as an outsider’s suggestion, I would try the following three step process

1. Search and Replace (non-regular-expression) <![CDATA[ to ∠∠, and ]]> to ⟂⟂ (where ∠∠ and ⟂⟂ were chosen as character sequences not likely in your document; if they are, pick some different delimiter
2. Since the CDATA markers are out of the way, XML Tools should validate and/or prettify the internal contents along with the structured
3. Search and Replace ∠∠ back to <![CDATA[, etc.

You might be able to get away with <was:CDATA> and </was:CDATA>, though I don’t know whether that would validate properly; I am not an XML expert.

Failing that, I would use a real XML tool to extract the CDATA; then run that extracted text through the prettify/validate; then repopulate the CDATA contents with the results. (less automatic, less fully contained in Notepad++/XML-Tools, but possibly easier to implement, if my three-step doesn’t work).

Personally, if the search/replace wasn’t sufficient, I would do that whole process in a single Perl script, using one of the XML modules in CPAN, like XML::LibXML or XML::Twig for the CDATA extraction and insertion, and piping through some external executable to get the prettification. Others here could probably recommend equivalent Python modules/packages/whatever-they’re-called. One benefit of a Python solution would be that you might be able to do it fully within the PythonScript plugin for Notepad++, so keeping it more contained.

Ooh, I just thought of something: Can XML Tools be told to just prettify or validate a particular selection, rather than the whole document? If so,

1. search for something like this regular-expression: (?s)<!\[CDATA\[\K.*?(?=\]\]>) (that appears to highlight the contents of an individual CDATA section
2. apply the XML Tools prettify or validate on just that selection
3. find-next and repeat on the next selection, ad infinitum

-----
update: by “regular expression”, I mean enable the ☑ Regular Expression option in the find or replace dialog window; by “non-regular-expression”, I mean enable the ☑ Normal option.